Journal of Applied Psychology1999, Vol. 84, No. 6, 885-896

Copyright 1999 by the American Psychological Association, Inc.0021-9010/99/S3.00

Goal Commitment and the Goal-Setting Process:Conceptual Clarification and Empirical Synthesis

Howard J. KleinOhio State University

Michael J. Wesson and John R. HollenbeckMichigan State University

Bradley J. AlgeOhio State University

Goals are central to current treatments of work motivation, and goal commitment is a criticalconstruct in understanding the relationship between goals and task performance. Despite thisimportance, there is confusion about the role of goal commitment and only recently has thiskey construct received the empirical attention it warrants. This meta-analysis, based on 83independent samples, updates the goal commitment literature by summarizing the accumu-lated evidence on the antecedents and consequences of goal commitment. Using this aggre-gate empirical evidence, the role of goal commitment in the goal-setting process is clarifiedand key areas for future research are identified.

Motivation continues to be a compelling topic for man-agers and researchers. In this vast literature, goals haveemerged as a central, pervasive construct (Austin & Van-couver, 1996). Goals are particularly integral among currenttheories of motivation that emphasize self-regulation. Theseperspectives include task goal theory (e.g., Locke &Latham, 1990), social-cognitive theory (e.g., Bandura,1986), resource allocation theory (e.g., Kanfer & Acker-man, 1989), and control theory (Klein, 1989). These theo-ries differ in several respects, but all feature goals as acentral determinant of motivation (Phillips, Hollenbeck, &Ilgen, 1996).

The bulk of the work examining goals as predictors ofperformance has been done within task goal theory. Thebasic finding from this theory is that under certain condi-tions, specific, difficult goals can lead to higher levels of

Howard J. Klein and Bradley J. Alge, Department of Manage-ment and Human Resources, Max M. Fisher College of Business,Ohio State University; Michael J. Wesson and John R. Hollenbeck,Department of Management, Eli Broad Graduate School of Man-agement, Michigan State University.

A version of this article was presented at the annual meeting ofthe Academy of Management, San Diego, California, August1998.

Correspondence concerning this article should be addressed toHoward J. Klein, Department of Management and Human Re-sources, Max M. Fisher College of Business, Ohio State Univer-sity, 2100 Neil Avenue, Columbus, Ohio 43210-1144. Electronicmail may be sent to klein.12@osu.edu.

performance relative to vague or easy goals (see Locke &Latham, 1990). One of the most often cited assumptions orconditions necessary for this relationship to hold is thatthere is commitment to that specific, difficult goal. Goalcommitment, one's determination to reach a goal (Locke &Latham, 1990), has been a central concept in goal-settingtheory since its inception.

In his 1968 article that launched the theory, as well as intwo subsequent reviews of the goal-setting literature(Locke, Shaw, Saari, & Latham, 1981; Locke & Latham,1990), Locke and his colleagues recognized that if there isno commitment, a goal can have no motivational effect.Although goal commitment has played a central conceptualrole in the development of task goal theory, the empiricalexamination of goal commitment was largely absent fromearly goal-setting research. Two separate reviews publishedin the late 1980s (Hollenbeck & Klein, 1987; Locke,Latham, & Erez, 1988) noted that the body of empiricalevidence on goal commitment's consequences and anteced-ents was insufficient given the variable's central role in thegoal-setting process. In fact, Hollenbeck and Klein con-cluded that the most frequent treatment of goal commitmentin the goal-setting literature was to not measure the variableat all but then to offer it up as an unvalidated, post hocexplanation whenever the goal difficulty effect was negli-gible, weak, or conditional on other variables.

Those reviews reiterated the important role of goal com-mitment, documented the absence of systematic researchinvestigating this crucial construct, and provided frame-works for guiding its investigation. Research in the area of

885

KLEIN, WESSON, HOLLENBECK, AND ALGE

goal setting has paid more attention to the construct of goalcommitment in the wake of those reviews. The purpose ofthis article is to report the results of a meta-analysis thatexamined the consequences and antecedents of goal com-mitment. This meta-analysis updates the goal commitmentliterature roughly 10 years after the Locke et al. (1988) andHollenbeck and Klein (1987) reviews and, using that ag-gregate empirical evidence, seeks to clarify the role of goalcommitment in the goal-setting process and identify keyareas for future research.

Consequences of Goal Commitment

The primary consequence of goal commitment is to mod-erate the relationship between goal difficulty and perfor-mance. There is considerable confusion in the literatureabout this relationship despite the fact that this assertion isconsistent with both early (Locke, 1968) and more recent(Locke & Latham, 1990) formulations of task goal theory.The confusion stems in part from the specific form of thismoderated relationship, technically referred to as an un-crossed interaction (Stone & Hollenbeck, 1984). As illus-trated in Figure 1, high performance comes about only whengoal difficulty and goal commitment are both high. Difficultgoals do not lead to high performance when commitment islow arid high levels of commitment to easy goals also fail togenerate high performance. Stated differently, a strong lin-ear relationship should be evident between goal difficultyand performance when commitment is high and goal diffi-culty should be unrelated to performance when commitmentis low. Because of the uncrossed nature of this interaction,main effects rather than the interaction can be expected

under certain operational conditions. In such situations, thefailure to observe a significant interaction does not mitigateor refute the critical role of goal commitment.

For example, the moderation hypothesis assumes thatthere is sufficient variance in both goal commitment andgoal difficulty. This is often not the case in goal-settingresearch or practice. Goal commitment cannot moderate therelationship between goal difficulty and performance ifthere is little variance in commitment. To function as amoderator, some individuals must have high commitment,some moderate commitment, and some low commitment. Insituations in which goals are self-set, goal commitment isgenerally high and invariant (Hollenbeck & Brief, 1987).Even when goals are assigned, situational demands (e.g.,incentives, competition, experimenter demand, legitimatepower) often result in uniformly high levels of commitment(Locke & Latham, 1990). Under such circumstances, a maineffect of goal difficulty on performance should be evidentand goal commitment should have negligible effects, mainor interactive (e.g., Klein, 1991; Matsui, Kakuyama, &Onglatco, 1987; Wright & Kacmar, 1995). Even when thereis variance in commitment, either within or across studies,there should be a main effect for goal difficulty on perfor-mance, averaged across all levels of commitment. Thedashed line in Figure 1 shows this average effect. This maineffect for goal difficulty on task performance across studieshas been documented elsewhere (e.g., Mento, Steel, & Kar-ren, 1987), and establishing this parameter was not a pur-pose of the present study.

Goal commitment also cannot moderate the relationshipbetween goal difficulty and performance if there is insuffi-

•§0>a..*(0re

High GoalCommitment

Average GoalCommitment

Low GoalCommitment

Low GoalDifficulty

Average GoalDifficulty

High GoalDifficulty

Figure 1. Conceptual interactive relationship between goal difficulty and goal commitment.

GOAL COMMITMENT 887

cient variance in goal difficulty levels. That is, some indi-viduals must have easy goals, some moderate goals, andothers difficult goals. Although this is the case in somesituations, all employees or participants are often assignedthe same challenging goal that violates this assumption. Insituations in which only challenging goals are present, com-mitment can be expected to have a main effect on perfor-mance (e.g., Harrison & Liska, 1994; Johnson & Perlow,1992; Klein & Kim, 1998). When everyone has the samedifficult goal, individuals with higher commitment to thatgoal should outperform those with lower commitment. Evenwhen there is variance in goal difficulty, either within oracross studies, this main effect for goal commitment onperformance should be evident, averaged across all levels ofgoal difficulty. The average effect for goal commitment isillustrated by the Xs corresponding to "average goal diffi-culty" in Figure 1. When goal difficulty levels are averaged,higher levels of commitment are associated with higherlevels of performance. Unlike the main effect of goal dif-ficulty, this overall main effect for goal commitment has notbeen established meta-analytically and thus serves as thefirst purpose of this meta-analysis.

Hypothesis 1: There will be a positive relationship betweengoal commitment and performance averaged across studies.

Although range restriction in either goal difficulty orcommitment is a common problem within studies, therehave been studies with sufficient variance in both goaldifficulty and goal commitment that have supported theinteractive effect (e.g., Erez & Zidon, 1984; Tubbs, 1993).Across the research that has been conducted, the interactivenature of the relationship depicted in Figure 1 should also beevident. Figure 1 suggests that low standings on eithervariable will negate the generally positive relationship be-tween performance and the other variable. Thus, the rela-tionship between commitment and performance will bestronger with difficult goals relative to easy goals. That is,goal difficulty moderates the relationship between commit-ment and performance. This moderating effect of goal dif-ficulty on the commitment-performance relationship hasnot been fully examined meta-analytically and thus servesas the second purpose of this meta-analysis.

Hypothesis 2: There will be a significant amount of variabil-ity in the goal commitment-performance relationship thatcannot be explained by sampling error or artifacts.

Hypothesis 3: Goal difficulty will moderate the relationshipbetween goal commitment and performance, such that therelationship between commitment and performance will bestronger for difficult goals relative to easy ones.

In the goal-setting literature, this interaction is typicallystated in terms of commitment moderating the goaldifficulty-performance relationship. Hypothesis 3 has goaldifficulty moderating the goal commitment-performance

relationship because goal difficulty levels can be catego-rized more objectively for conducting a meta-analytic mod-erator analysis than levels of goal commitment. Becauseinteractive relationships are symmetrical in nature (Cohen& Cohen, 1983), the conclusion that goal difficulty moder-ates the goal commitment-performance relationship neces-sarily substantiates the idea that goal commitment moder-ates the goal difficulty-performance relationships. Thus, theway we used meta-analysis to test for moderation wasindirect, but the overall conclusion remained the same: Highlevels of both commitment and goal difficulty are necessaryto achieve high levels of performance.

An alternative approach to meta-analytically test the in-teraction hypothesis would be to directly aggregate theeffect sizes for this interaction reported in the literature(e.g., Donovan & Radosevich, 1998). Such an approach isproblematic because of serious limitations in the data avail-able in the literature. First, only a handful of publishedstudies have reported directly testing the interaction hypoth-esis using multiple regression. Second, for some of thosestudies, an interaction would not be expected because of theproblem of restricted variance discussed earlier. For exam-ple, Donovan and Rodosevich (1998) included Harrison andLiska (1994) in their analysis even though goals were heldconstant in that study. Third, some studies presenting strongevidence of the interaction effect (e.g., Erez & Zidon, 1984)did not use a multiple regression approach and could not beincluded. Also excluded from such an approach are thedozens of published studies that examined goals, commit-ment, and performance but did not report any moderatoranalyses. As a result, Donovan and Radosevich's analyseswere limited to data from only 6 published articles, whereasthe current meta-analytic test for moderation was based ondata from more than 60 published articles. To directly testthe interaction hypothesis based on a representative sampleof all studies measuring goals, commitment, and perfor-mance, one would need to have the study intercorrelationsbetween the main effect variables and their cross-products.Such statistics do not exist in the literature.

Antecedents of Goal Commitment

If goal commitment has important performance conse-quences, as predicted, then attention must also be directed atfactors that affect goal commitment. The final purpose ofthis meta-analysis is to examine antecedents of goal com-mitment. Hollenbeck and Klein (1987), consistent withLocke et al. (1981), used an expectancy theory frameworkin delineating the determinants of goal commitment. Theattractiveness of goal attainment and the expectancy of goalattainment are thus thought to be the most proximal ante-cedents of goal commitment. Across the accumulated liter-ature, those two variables along with motivational force, themultiplicative combination of expectancy and attractive-

KLEIN, WESSON, HOLLENBECK, AND ALGE

ness, should be strongly related to goal commitment. Al-though previous meta-analyses (e.g., Klein, 1991; Wofford,Goodwin, & Premack, 1992) have provided support forthese relationships, in the current study we used a substan-tially larger sample of studies than earlier efforts.

Hypothesis 4: There will be positive relationships betweengoal commitment and expectancy, attractiveness, and moti-vational force averaged across studies.

Although expectancy and attractiveness were the mostproximal antecedents of goal commitment in Hollenbeckand Klein's (1987) model, several other variables wereidentified as likely to affect goal commitment through theireffects on the expectancy and attractiveness of goal attain-ment. Empirical evidence exists for some of these relation-ships but not for others. Furthermore, the variables includedin Hollenbeck and Klein's model were provided to beillustrative rather than a comprehensive or exhaustive set ofantecedents. Although the additional antecedents of goalcommitment suggested in that model and other past reviewsshould also relate to goal commitment, we offer no formalhypotheses with respect to all of these variables. We sum-marize the findings in the literature for all other correlates ofgoal commitment, whether explicitly identified in a previ-ous re.view or not.

Method

Literature Search

Although there has been a recent surge in research on goalcommitment, we did not limit this meta-analysis to those latestfindings. Studies included in prior reviews were also incorporatedhere to provide a comprehensive analysis. We conducted a searchfor studies reporting empirical findings involving goal commit-ment. Both computer and manual methods were used in identifyingrelevant studies to obtain the largest sample possible. First, com-puterized searches were conducted through both the PsycLIT andABI/Inform databases using the key words goal commitment, goalacceptance, and goal difficulty. Second, previous reviews of thegoal commitment literature (Hollenbeck & Klein, 1987; Locke &Latham, 1990; Locke et al., 1988; Wofford et al., 1992) wereexamined. Next, we manually searched the Academy of Manage-ment Journal, Human Performance, Journal of Applied Psychol-ogy, Journal of Applied Social Psychology, Journal of Manage-ment, Organizational Behavior and Human Decision Processes,and Personnel Psychology from January 1987 through May 1998.We chose these journals because they were most likely to containgoal-setting studies given the distribution of articles identifiedthrough the previous steps. Studies published before 1987 shouldhave been included in the previous reviews of goal commitment.Finally, the reference lists of studies identified using these proce-dures were examined for any additional studies.

This search of the literature identified 174 published articles,dissertations, book chapters, and conference papers. From these,the following criteria for inclusion were used for the currentmeta-analyses: (a) Goal commitment or goal acceptance had to be

measured in the study and (b) suitable statistics had to be presentto calculate an effect size between goal commitment and at leastone other construct. The application of these inclusion criteriaresulted in the identification of 74 studies. Nine of these studiesincluded more than one sample or study, bringing the total to 83independent samples. Sample sizes ranged from 20 to 406 (M =105.06, SD = 72.11). The studies included in the meta-analysesare indicated by an asterisk in the reference list. Note that morethan half (n = 48) of these studies were published after the reviewsby Hollenbeck and Klein (1987) and Locke et al. (1988).

Coding of Studies and Constructs

Each study was coded for sample size, effects between theconstructs of interest, and measurement reliability. Two of theauthors each coded half of the studies, with 10 studies coded byboth authors. The agreement between the raters was 95% on thosestudies. All disagreements were resolved through discussion withthe first author. In addition, information was collected on eachstudy regarding several potential moderators: (a) task complexity(using the categories provided by Wood, Mento, & Locke, 1987);(b) incentives (yes vs. no); and (c) goal origin (self-set vs. partic-ipative vs. assigned). These variables were identified as potentialmoderators based on previous goal-setting meta-analyses (Mentoet al., 1987; Wofford et al., 1992; Wood et al., 1987). Given thesmall number of studies identified using participatively set goals,we focused the goal origin analyses only on self-set versus as-signed goals.

Goal difficulty was also coded as a moderator to test Hypoth-esis 3. In doing so, studies (or conditions within studies) werecategorized as using easy, moderate, or difficult goals. That cate-gorization was based on the objective probability of goal attain-ment when possible. Goals reported or judged as having an ob-jective probability of attainment of less than 15% were coded asdifficult1; goals between 16% and 50% were coded as moderate;and goals reported or judged as having an objective probability ofattainment greater than 50% were coded as easy. The coding of theother constructs examined was relatively straightforward. In sev-eral cases, similar constructs were combined in the interest ofparsimony. We further explain those variables.

Goal commitment. Although conceptual distinctions can bemade between the constructs of goal acceptance and goal commit-ment, studies that examined one or the other are included hereunder the term goal commitment. This is consistent with previousreviews (e.g., Hollenbeck & Klein, 1987; Locke & Latham, 1990)and the emerging consensus that commitment is the more inclusiveconstruct. All self-report measures of goal commitment were in-cluded in this analysis, even those using single-item measures.Excluded were studies using only discrepancy measures of goalcommitment (see Wright, O'Leary-Kelly, Cortina, Klein, & Hol-

1 In experimental studies assigning goals of varying difficultylevels, a "difficult" goal has often been defined as having anobjective probability of attainment of 10%. For the purposes of thisreview, using a cutoff of 10% in categorizing goal difficulty wasviewed as overly restrictive. Our use of 15% in this review shouldnot be viewed as a recommended operationalization of difficultgoals in future experimental manipulations.

GOAL COMMITMENT 889

lenbeck, 1994) or studies in which participants evaluated thecommitment of others.

Performance. For the majority of studies, objective perfor-mance indexes were available to use as the measure of taskperformance. When both objective and subjective measures wereavailable, we used the objective measure. In instances in whichmultiple dimensions of performance were evaluated, we chose themeasure closest to the goal of interest as the performance measure(in most cases, this was the quantity of work).

Expectancy. Studies examining a variety of expectancy con-structs (e.g., outcome expectancies, effort-performance expectan-cies) were included in this category as were self-efficacy expec-tations. Although conceptually distinct, nearly identical measureshave been used to assess self-efficacy and expectancy in thegoal-setting literature (Klein, 1991).

Attractiveness. Direct measures of the attractiveness of goal at-tainment were grouped with measures of valance or instrumentality.

Goal level. The performance level specified in the goal forwhich commitment was assessed was used as the measure of goaldifficulty level. If, for example, a study assigned goals and as-sessed commitment to those assigned goals but later also assessedpersonal goals, the assigned goal level was used in the meta-analysis. In studies in which commitment was assessed relative topersonal goals, the difficulty of those self-set goals was used.

Ability. Measures of ability were combined with the com-monly used surrogates for ability, namely previous or practiceperformance. When measures of both ability and past performancewere available, we used the ability measure.

Feedback. Two groupings of measures were created concern-ing feedback or knowledge of results. The first, provision offeedback, included studies comparing feedback with no-feedbackconditions as well as those measuring or manipulating the amountof feedback provided. The second set of studies, type of feedback,included studies that measured or manipulated the nature or someother dimension of the feedback received by study participants.

Meta-Analytic Procedures

Meta-analyses were conducted using the procedures outlined byHunter and Schmidt (1990). Their procedures examined the degreeto which correlation differences across studies could be accountedfor by statistical artifacts (e.g., sampling error and unreliability)and allowed for the correction of these correlations to obtain the

true population correlation. Population correlations were estimatedfor each of the relationships using the observed correlationsweighted for sample size and the calculated variance of correla-tions across samples. Each sample-weighted correlation was cor-rected for measurement error in goal commitment. Reliabilityinformation was presented for more than 75% of the goal com-mitment measurements. The average reliability for goal commit-ment was .81. When unreported, this mean was used for correctionpurposes. We did not correct for measurement error in goal diffi-culty, performance, or the coded antecedents because most of thesemeasures were objective in nature, were experimentally manipu-lated, or the reliabilities were unreported in the literature. Themoderating effects discussed in this study were analyzed using theguidelines set by Hunter and Schmidt (1990) and Whitener (1990).We examined the percentage of variance explained by artifactsaccording to Hunter and Schmidt's 75% rule and homogeneitysignificance tests; we also examined the credibility intervals forthe degree of spread and the inclusion of zero.

Results and Discussion

Consequences of Goal Commitment

Hypotheses 1-3 concerned the relationship between goalcommitment and performance. The meta-analytic results forthis relationship are provided in Table 1. The mean cor-rected effect size (rc) between goal commitment and per-formance was .23. The 95% confidence interval (CI) for thisweighted mean correlation did not include zero, supportingHypothesis 1, which predicted a positive relationship be-tween goal commitment and performance across studies.Hypothesis 2 predicted a significant amount of variability inthis relationship that would not be explained by samplingerror or artifacts. As indicated in Table 1, only 27% of thevariance in results across studies was attributable to sam-pling error. The homogeneity chi-square was significant atthe .01 level, indicating support for Hypothesis 2 and a highprobability that moderators of this relationship werepresent.

Hypothesis 3 predicted that goal difficulty would mod-erate this relationship. The results of the moderation anal-

Table IMeta-Analysis Results of the Goal Commitment-Task Performance Relationship

Goal difficultylevel

HighModerateLowOverall

k

17221666

N

1,2933,0931,5317,952

r

.31

.18

.16

.20

Lower95% CI

.25

.12

.10

.16

Upper95% CI

.38

.24

.22

.25

% due tosampling error

52.8829.7771.9426.77

. X2

32.15*73.90**22.24

246.53**

rc.

.35

.20

.18

.23

Note. Significant correlations are those whose confidence intervals do not include zero, k = number ofcorrelations; N = total number of individuals across the k samples; r = mean sample weighted correlation; 95%CI = 95% confidence interval of the weighted mean correlation; % due to sampling error = percentage ofvariance explained by sampling error alone; x* = homogeneity of effect sizes within each class; rc = meanweighted correlation corrected for goal commitment unreliability.* p < .05. **p < .01.

890 KLEIN, WESSON, HOLLENBECK, AND ALGE

ysis, also presented in Table 1, reflect the predicted patternillustrated in Figure 1. The relationship between commit-ment and performance was stronger for difficult goals rel-ative to easy goals, supporting Hypothesis 3. The correctedaverage correlation for difficult goals (rc = .35) was sig-nificantly higher (p < .05) than the average correlation formoderate (rc = .20) or low (rc = .18) goals, as evidenced bythe fact that the 95% CI for difficult goals did not overlapwith the CIs for moderate or low goals. Further support forthe moderating effect was evident in the increased varianceaccounted for by sampling error for the subsets relative tothe overall effect. The homogeneity chi-square was nonsig-nificant for the low goal-difficulty category but significantfor the high and moderate subsets, suggesting the presenceof additional moderators.

Other potential moderators (e.g., goal origin, task com-plexity, incentives) of the goal commitment-performancerelationship were examined but not supported. Secondarymoderators within levels of goal difficulty were not exam-ined because of the diminished number of studies in eachsubset. Although all of the sampling variance was not ac-counted for, strong support was found for Hypotheses 1-3.There was a strong cross-situational positive relationshipbetween goal commitment and performance. In addition,moderator analyses showed that the effect of goal commit-ment on performance was considerably stronger in situa-tions in which difficult goals were used. Given the previ-ously established robust relationship between goal difficultyand task performance (Locke & Latham, 1990; Mento et al.,1987), these findings highlight the importance of facilitatingcommitment to difficult goals.

Antecedents of Goal Commitment

Hypothesis 4 predicted positive relationships betweengoal commitment and expectancy, attractiveness, and mo-tivational force. The meta-analytic results for these relation-ships are provided in Table 2. All three weighted meancorrelations had CIs that placed them significantly differentfrom zero, supporting the hypothesis. For expectancy, the

mean effect size, corrected for measurement error (rc), was.36. The weighted mean effects were .29 for attractivenessand .33 for motivational force. For all three relationships,substantial variance in results across studies remained afteraccounting for sampling error and the homogeneity chi-squares were significant at the .01 level. Although not afocus of the current study, different operationalizations ofthese variables (e.g., summated vs. single value expectan-cies) have been shown to moderate these relationships(Klein, 1991). Although there may be additional modera-tors, Hypothesis 4 was supported in that there were strongpositive overall relationships between goal commitment andattractiveness of goal attainment, expectancy of goal attain-ment, and motivational force.

We did not formulate any specific hypotheses about thenumerous distal antecedents of goal commitment. Themeta-analytic results for these other correlates of commit-ment are provided in Table 3, organized by the amount ofevidence available for each antecedent. There are relativelyfew additional variables that have been extensively studied.Of these, significant positive relationships were found forability (rc = .18), volition (rc = .40), affect (rc = .22), goalspecificity (rc = .19), task experience (rc = .24), and boththe provision (rc = .13) and type (rc = .30) of feedback.Higher levels of commitment resulted from having highability, a voice in the determination of the goal, task or jobsatisfaction, specific goals, task experience, receiving feed-back on one's performance, and the form of that feedback.Except for the provision of feedback, substantial varianceacross studies remained after accounting for sampling error,and the homogeneity chi-squares were significant. A signif-icant relationship could not be inferred for age, which wasnot surprising because there is no conceptual basis for sucha relationship.

Goal difficulty was the most frequently examined corre-late of goal commitment. Conceptually, the literature hasfocused on the interaction between goal difficulty and com-mitment in determining performance rather than on therelationship between goal difficulty level and goal commit-

Table 2Meta-Analysis Results for the Proximal Antecedents of Goal Commitment

Variable

ExpectancyAttractivenessMotivational force

k

24134

N

2,9952,075

593

r

.32

.26

.30

Lower95% CI

.26

.18

.06

Upper95% CI

.39

.34

.54

% due tosampling error

24.9019.419.16

x2

95.89**54.34**43.71**

rc.

.36

.29

.33

Note. Significant correlations are those whose confidence intervals do not include zero, k = number ofcorrelations; N = total number of individuals across the k samples; r = mean sample weighted correlation. 95%CI = 95% confidence interval of the weighted mean correlation; % due to sampling error = percentage ofvariance explained by sampling error alone; x2 = homogeneity of effect sizes within each class; rc = meanweighted correlation corrected for goal commitment unreliability.**p < .01.

GOAL COMMITMENT 891

Table 3Meta-Analysis Results for the Distal Antecedents of Goal Commitment

Variable

Goal levelAbility/past performanceVolition (participation)AffectExplicitness (goal specificity)Task information/experienceProvision of feedbackType of feedbackAgeTask complexityStrategy developmentIncentivesNeed for achievementLeadershipEducationSocial influenceSupervisor supportivenessPublicnessType A personalityStressGoal stabilityEndurance (persistence)Procedural justicePerformance constraintsGoal conflictOrganizational commitmentLocus of controlPerfectionismSelf-esteemAuthorityCompetitionJob involvement

k

40191777665533333322222111111111000

N

4,8922,4312,007

75096856577240451918632034032424334320893

248499252

3918418645

13275

19026188

r

.03

.16

.37

.20

.17

.22

.12

.27

.08-.50

.21

.20

.17

.12

.04

.45

.38

.19

.12-.10

.42

.30

.29-.27

.24

.20-.18

.07-.03

Lower95% CI

-.05.08.28.11.04.06.04.12

-.02-.64

.11

.10

.01

.00-.07

.32

.20

.07

.03-.36

Upper95% CI

.11

.24

.46

.29

.31

.38

.21

.42

.18-.35

.32

.31

.34

.25

.14

.58

.55

.31

.20

.17

% due tosampling error

12.7022.8516.5753.4221.1225.8470.7836.8278.0353.90

100.00100.0040.73

100.00100.0070.14

100.00100.00100.0021.75

X2

314.91**83.12**

102.22**13.12*33.03**23.21**

8.4713.61**6.405.571.112.667.37*0.361.642.850.790.930.369.19**

rc

.04

.18

.40

.22

.19

.24

.13

.30

.09

Note. Significant correlations are those whose confidence intervals do not include zero. Variables in italics arethose identified in previous reviews by Hollenbeck and Klein (1987) and Locke, Latham, and Erez (1988). k =number of correlations; N = total number of individuals across the k samples; r = mean sample weightedcorrelation. 95% CI = 95% confidence interval of the weighted mean correlation; % due to sampling error =percentage of variance explained by sampling error alone; )f = homogeneity of effect sizes within each class;rc = mean weighted correlation corrected for goal commitment unreliability.* p < .05. ** p < .01.

ment. In fact, neither Hollenbeck and Klein (1987) norLocke et al. (1988) included goal difficulty in their discus-sions of the antecedents of goal commitment. At a simplelevel, one might expect goal commitment to decline as goalsbecome objectively more difficult. This relationship is morecomplex than that, however, and the corrected averagecorrelation reported in Table 3 is close to zero (rc = .03).The low correlation and the wide variance observed acrossstudies indicates a strong possibility for moderators. Goalorigin was examined as a potential moderator based on thefindings by Wofford et al. (1992), who examined personaland assigned goals separately. In the current analysis, theaverage weighted effect (rc) was —.09 for assigned goals(95% CI = -.19 to .03; 12.06% of variance due to sampleerror), *2(18, N = 2,424) = 157.61, p < .01. For self-setgoals, there was an overall positive relationship (rc = .16)that was significantly different from zero based on the CI

(.06 to .25; 25.16% of variance due to sample error),N = 1,630) = 59.63, p < .01. For both subsets, the signif-icant homogeneity statistics indicated that goal origin alonedid not explain the variability in sample effects.

Consideration of the more proximal antecedents of com-mitment may help explain the relationship between goaldifficulty level and goal commitment and the moderatingrole of goal origin. Although the objective probability ofattainment becomes lower as goals become more difficult,this is not necessarily reflected in individual expectations ofgoal attainment (i.e., self-efficacy can remain high). Inaddition, although objectively difficult goals may havelower probabilities of attainment, they also often havehigher instrumentalities (Matsui et al., 1981; Mento, Locke,& Klein, 1992). Perhaps there tends to be high commitmentto difficult personal goals because those self-chosen goalstend to have high instrumentalities and expectancies higher

892 KLEIN, WESSON, HOLLENBECK, AND ALGE

than warranted by their objective difficulty. For assignedgoals, there is likely to be much more variability in both theexpectancy and attractiveness of goal attainment because ofother individual and situational factors. Future researchersneed to directly examine this suggested role of goal origin,expectancies, and attractiveness in understanding the goaldifficulty-goal commitment relationship.

Other antecedents, although not studied extensively, haveat least been replicated, allowing for the calculation of anaverage correlation. Because these effects were based on asmall number of independent samples, these results must beevaluated with caution. Several other variables have beenexamined only once. The effects for these variables shouldnot be interpreted as estimates of population correlations.They are included in Table 3 only to provide a comprehen-sive summary of the variables that have been examined aspotential antecedents of goal commitment and to identifyrelationships in need of replication. Also listed in Table 3are authority, competition, and job involvement. These vari-ables were identified in previous reviews as potential ante-cedents of goal commitment but have not been examinedempirically.

Study Limitations

By their very nature, meta-analyses call for many subjec-tive decisions (Wanous, Sullivan, & Malinak, 1989). Forexample, decisions were made here regarding study inclu-sion, significance thresholds, and classification rules. Thesedecisions were made as objectively as possible, but thepotential for bias needs to be recognized. A second limita-tion is the heterogeneity in many overall effect sizes, sug-gesting that the simple weighted mean correlations do notadequately describe all sampling variance. Although weconducted a few moderator analyses, some of which helpedto reduce this heterogeneity, the requirements for completehomogeneity were often not achieved. For some antecedentvariables, this heterogeneity is likely attributable to thevariety of constructs or operationalizations that were com-bined into a single category (e.g., feedback type).

Another limitation was the inability to conduct a hierar-chical breakdown of moderators because of the limitednumber of studies in each subset. Problems with statisticalpower and meta-analysis have been discussed elsewhere(Hunter & Schmidt, 1990; Sackett, Harris, & Orr, 1986). Itwould also have been useful to conduct a meta-analytic pathanalysis (e.g., Horn, Caranikas-Walker, Prussia, & Griffeth,1992). Unfortunately, we could not find sufficient data inthe literature to generate a complete meta-analytic correla-tion matrix among the variables of interest. Another limita-tion is the lack of consistency with which the goal commit-ment construct has been operationalized, which makes itdifficult to argue that all included studies were measuringthe exact same thing (see Tubbs & Dahl, 1991, and Wright

et al., 1994). Although we limited the operationalizations,we included a variety of measures. Finally, as mentionedearlier, the data available in the literature and the fact thatvariation in goal commitment is often limited, as is thevariation in goal level, made it inappropriate to use meta-analysis to directly examine goal commitment as a moder-ator in the goal difficulty-performance relationship.

Future Research Needs

Hollenbeck and Klein (1987) noted that goal commitmentwas not measured or mentioned in the majority of goal-setting studies they reviewed. They and Locke and Latham(1990) recommended that goal commitment be measured inall goal-setting studies, even if goal commitment is not avariable of interest. By doing so, goal commitment serves asa manipulation check and can be used to test the most likelyexplanation if hypothesized goal effects are not observed. Itappears that recent researchers have largely followed thisadvice, although we did not attempt to identify goal-settingstudies not assessing goal commitment. This is evidencedby the large number of studies identified in our literaturesearch and not included in this meta-analysis because theonly statistics reported concerning goal commitment wereto convey that all participants were committed to the goalsbeing investigated.

A disadvantage of this practice is that goal commitment isoften a secondary variable. That is, commitment is mea-sured without specific hypotheses aimed at exploring itsnature and role, and statistical relationships are not reportedbetween goal commitment and the other variables beingstudied. Furthermore, when goal commitment is used as amanipulation check, steps are often taken to ensure high andinvariant levels of commitment instead of taking steps tomaximize variability in order to observe relationships withgoal commitment. As a result, although goal commitmenthas been included in a greater percentage of goal-settingstudies, what researchers have learned about goal commit-ment is not fully reflected by that attention. This lack ofattention to goal commitment as a primary variable ofinterest is best exemplified by the relatively few studies thathave specifically and appropriately examined the interactiveeffect of goal commitment and goal difficulty on perfor-mance. Thus, a first need is for additional research to verifydirectly what this meta-analysis has shown indirectly: thatwith sufficient variance in both goal commitment and goaldifficulty levels, goal commitment moderates the relation-ship between goal difficulty and task performance.

The current meta-analysis, consistent with Hollenbeckand Klein's (1987) model, showed that the expectancy andattractiveness of goal attainment were highly related to goalcommitment. Although several other variables were alsofound to relate to goal commitment, in this meta-analysis wecould not test the mediating role of expectancy and attrac-

GOAL COMMITMENT 893

tiveness implicit in Hollenbeck and Klein's model. In ad-dition, relatively few researchers have directly examinedthat relationship (e.g., Klein & Wright, 1994; Wright &Kacmar, 1995). Further examination of this linkage is asecond area for future research. A third area for futureresearch concerns the examination of previously identifiedantecedents that have been infrequently examined (e.g.,goal conflict, performance constraints) or not examined atall (e.g., authority, competition). In addition, there may beother variables (e.g., conscientiousness, goal orientation)that, although not previously identified, are worthy of in-vestigation as antecedents of goal commitment.

A strength of the recent goal commitment literature is thedefinition and measurement of the construct. In particular,the use of a variety of single-item measures has given wayto the use of multi-item, standardized, reliable instruments.Hollenbeck, Williams, and Klein's (1989) scale and itsderivatives are the most commonly used measures of goalcommitment. There has also been greater convergence be-tween the use of the terms goal acceptance and goal com-mitment. Whereas previously the two terms were the sourceof much confusion and were used interchangeably, goalcommitment has emerged as the more inclusive of the twoconstructs and has received the bulk of attention in recentyears. Despite this convergence on the meaning and mea-surement of goal commitment, there is still some debateover both of these issues (DeShon & Landis, 1997; Tubbs,1993). Empirical studies clarifying these remaining pointsof contention represent another future research need.

Much of the goal commitment literature has assumed thathigh commitment is desirable. The escalation of commit-ment literature (e.g., Staw, 1982) clearly suggests, however,that there are situations in which high commitment is dys-functional. It is also conceivable that there are situations inwhich excessively high commitment is detrimental to indi-vidual well-being because of the stress, anxiety, or otherhealth risks caused by relentless goal striving (e.g., worka-holism). Research examining potential negative implica-tions of goal commitment is thus warranted. A final impor-tant area for future research is the examination ofcommitment to multiple goals. An individual's commitmentto a particular goal is most meaningful relative to that sameperson's commitment to other goals. One may be highlycommitted to a work-related goal, but if that person is evenmore committed to a conflicting family-related goal, highcommitment to the work goal will not be predictive ofperformance. Few researchers have examined commitmentto multiple goals (e.g., Crown & Rosse, 1995; Locke,Smith, Erez, Chah, & Schaffer, 1994).

Conclusion

On the basis of a small sample of studies, Donovan andRadosevich (1998) concluded that the evidence in the liter-

ature did not support the importance or hypothesized role ofgoal commitment in the goal-setting process. Results of thecurrent study, using a larger and more representative sampleof studies, refutes those assertions. The main findings of thisreview are as follows: (a) Goal commitment had a strongpositive effect on performance across studies; (b) goal dif-ficulty moderated this relationship (and hence goal commit-ment moderated the goal difficulty-performance relation-ship); (c) expectancy and attractiveness of goal attainment,the antecedents thought to be most proximal, were stronglyand positively related to goal commitment; and (d) only ahandful of relationships were shown to have been welldocumented in the literature between goal commitment andmore distal antecedents (e.g., ability, volition).

Goals are central to current treatments of work motiva-tion, and goal commitment is a necessary condition fordifficult goals to result in higher task performance. A de-cade ago, the call was made for more systematic researchinvestigating this crucial construct. The number of recentstudies identified in this meta-analysis suggests that goalcommitment has received increased attention. However, thelimited number of studies directly and appropriately exam-ining the interaction between goal commitment and perfor-mance, the conceptual confusion that remains about thisrelationship, and the sparse attention given to several pre-viously identified antecedent variables suggest that thisresearch has not been sufficiently systematic. We hope thatthis conceptual clarification and empirical synthesis willfacilitate more systematic research in the next decade.

References

References marked with an asterisk indicate studies in-cluded in the meta-analysis.*Allscheid, S. P., & Cellar, D. F. (1996). An interactive approach

to work motivation: The effects of competition, rewards, andgoal difficulty on task performance. Journal of Business andPsychology, 11, 219-237.

*Anderson, J. C., & O'Reilly, C. A. (1981). Effects of an organi-zational control system on managerial satisfaction and perfor-mance. Human Relations, 34, 491-501.

*Austin, J. T. (1989). Effects of shifts in goal origin on goalacceptance and attainment. Organizational Behavior and Hu-man Decision Processes, 44, 415-435.

Austin, J. T., & Vancouver, J. B. (1996). Goal constructs inpsychology: Process and content. Psychological Bulletin, 120,

338-375.Bandura, A. (1986). Social foundations of thought and action: A

social-cognitive view. Englewood Cliffs, NJ: Prentice Hall.*Brunstein, J. C. (1993). Personal goals and subjective well-being:

A longitudinal study. Journal of Personality and Social Psy-chology, 65, 1061-1070.

Cohen, J., & Cohen, P. (1983). Applied multiple regression/cor-relation analysis for the behavioral sciences (2nd ed.). Hills-dale, NJ: Erlbaum.

Crown, D. F., & Rosse, J. G. (1995). Yours, mine, and ours:

894 KLEIN, WESSON, HOLLENBECK, AND ALGE

Facilitating group productivity through the integration of indi-vidual and group goals. Organizational Behavior and HumanDecision Processes, 64, 138-150.

*DeShon, R., & Alexander, R. (1996). Goal setting effects onimplicit and explicit learning of complex tasks. OrganizationalBehavior and Human Decision Processes, 65, 18-36.

DeShon, R., & Landis, R. S. (1997). The dimensionality of theHollenbeck, Williams, and Klein (1989) measure of goal com-mitment on complex tasks. Organizational Behavior and Hu-man Decision Processes, 70, 105-116.

*De Souza, G., & Klein, H. J. (1995). Emergent leadership in thegroup goal-setting process. Small Group Research, 26, 475-496.

*Dodd, N. G., & Anderson, K. S. (1996). A test of goal commit-ment as a moderator of the relationship between goal level andperformance. Journal of Social Behavior and Personality, 11,329-336.

Donovan, J. J., & Radosevich, D. J. (1998). The moderating role ofgoal commitment on the goal difficulty-performance relation-ship: A meta-analytic review and critical reanalysis. Journal ofApplied Psychology, 83, 308-315.

*Durham, C. C., Knight, D., & Locke, E. A. (1997). Effects ofleader role, team-set goal difficulty, efficacy, and tactics onteam effectiveness. Organizational Behavior and Human Deci-sion Processes, 72, 203-231.

*Earley, P. C. (1985a). Influence of information, choice and taskcomplexity upon goal acceptance, performance, and personalgoals. Journal of Applied Psychology, 70, 481-491.

*Earley, P. C. (1985b, August). The influence of goal settingmethods on performance, goal acceptance, self-efficacy expec-tations, and expectancies across levels of goal difficulty. Paperpresented at the 93rd Annual Convention of the AmericanPsychological Association, Los Angeles.

*Earley, P. C., & Kanfer, R. (1985). The influence of componentparticipation and role models on goal acceptance, goal satisfac-tion, and performance. Organizational Behavior and HumanDecision Processes, 36, 378-390.

*Early, P. C., Shalley, C. E., & Northcraft, G. B. (1992). I think Ican, I think I can: Processing time and strategy effects of goalacceptance/rejection decisions. Organizational Behavior andHuman Decision Processes, 53, 1-13.

*Erez, M., & Arad, R. (1986). Participative goal setting: Social,motivational, and cognitive factors. Journal of Applied Psychol-ogy, 71, 591-597.

*Erez, M., Barley, C. P., & Hulin, C. L. (1985). The impact ofparticipation on goal acceptance and performance: A two-stepmodel. Academy of Management Journal, 28, 50-66.

*Erez, M., & Somech, A. (1996). Is group productivity loss therule or the exception? Effects of culture and group-based moti-vation. Academy of Management Journal, 39, 1513-1537.

Erez, M., & Zidon, I. (1984). Effect of goal acceptance on therelationship of goal difficulty to performance. Journal of Ap-plied Psychology, 69, 69-78.

*Flett, G. L., Sawatzky, D. L., & Hewitt, P. L. (1995). Dimensionsof perfectionism and goal commitment: A further comparison oftwo perfectionism measures. Journal of Psychopathology andBehavioral Assessment, 17, 111-124.

*French, J. P. R., Kay, E., & Meyer, H. H. (1966). Participationand the appraisal system. Human Relations, 19, 3-19.

*Guhde, L. M. (1991). The influence of task complexity, strategyand group goals on performance and goal acceptance. Unpub-lished doctoral dissertation, Kent State University, Kent, OH.

*Hanges, P. J. (1986). A catastrophe model of control theory'sdecision mechanism: The effects of goal difficulty, task diffi-culty, goal direction and task direction on goal commitment.Unpublished doctoral dissertation, University of Akron, Akron,OH.

*Hanges, P. J., Alexander, R. A., & Herbert, G. R. (1987, April).Using regression analysis to empirically verify catastrophemodels. Paper presented at the annual Society of Industrial andOrganizational Psychology meeting, Atlanta, GA.

*Harrison, D. A., & Liska, L. Z. (1994). Promoting regular exer-cise in organizational fitness programs: Health related differ-ences in motivational building blocks. Personnel Psychol-ogy, 47, 47-71.

*Hinsz, V. B., Kalnbach, L. R., & Lorentz, N. R. (1997). Usingjudgmental anchors to establish challenging self-set goals with-out jeopardizing commitment. Organizational Behavior andHuman Decision Processes, 71, 287-308.

Hollenbeck, J. R., & Brief, A. P. (1987). The effects of individualdifferences and goal origin on the goal setting process. Orga-nizational Behavior and Human Decision Processes, 40, 392-414.

Hollenbeck, J. R., & Klein, H. J. (1987). Goal commitment and thegoal-setting process: Problems, prospects, and proposals forfuture research. Journal of Applied Psychology, 72, 212-220.

*Hollenbeck, J. R., Williams, C. L., & Klein, H. J. (1989). Anempirical examination of the antecedents of commitment todifficult goals. Journal of Applied Psychology, 74, 18-23.

Horn, P. W., Caranikas-Walker, F., Prussia, G. E., & Griffeth,R. W. (1992). A meta-analytical structural equations analysis ofa model of employee turnover. Journal of Applied Psychol-ogy, 77, 890-909.

*Huber, V. L., & Neale, M. A. (1986). Effects of cognitiveheuristics and goals on negotiator performance and subsequentgoal setting. Organizational Behavior and Human DecisionProcesses, 38, 342-365.

Hunter, J. E., & Schmidt, G. L. (1990). Methods ofmeta-analysis:Correcting error and bias in research findings. Newbury Park,CA: Sage.

*Ivancevich, J. M., & McMahon, J. T. (1977a). Education as amoderator of goal setting effectiveness. Journal of VocationalBehavior, 11, 3-94.

*Ivancevich, J. M., & McMahon, J. T. (1977b). Black-whitedifferences in a goal-setting program. Organizational Behaviorand Human Performance, 20, 287-300.

*Ivancevich, J. M., & McMahon, J. T. (1977c). A study of task-goal attributes, higher order need strength, and performance.Academy of Management Journal, 20, 552-563.

*Johnson, D. S., & Perlow, R. (1992). The impact of need forachievement components on goal commitment and perfor-mance. Journal of Applied Social Psychology, 22, 1711-1720.

Kanfer, R., & Ackerman, P. L. (1989). Motivation and cognitiveabilities: An integrative/aptitude-treatment approach to skill ac-quisition. Journal of Applied Psychology, 74, 657-690.

GOAL COMMITMENT 895

*Kernan, M. C., & Lord, R. G. (1988). Effects of participative vs.assigned goals and feedback in a multi-trial task. Motivation &Emotion, 12, 75-86.

Klein, H. J. (1989). An integrated control theory model of workmotivation. Academy of Management Review, 14, 150-172.

*Klein, H. J. (1991). Further evidence on the relationship betweengoal setting and expectancy theories. Organizational Behaviorand Human Decision Processes, 49, 230-257.

*Klein, H. J., & Kim, J. S. (1998). A field study of the influenceof situational constraints, leader-member exchange, and goalcommitment on performance. Academy of Management Jour-nal, 41, 88-95.

*Klein, H. J., & Mulvey, P. W. (1995). Two investigations of therelationships among group goals, goal commitment, cohesion,and performance. Organizational Behavior and Human Deci-sion Processes, 61, 44-53.

*Klein, H. J., & Wright, P. M. (1994). Antecedents of goalcommitment: An empirical examination of personal and situa-tional factors. Journal of Applied Social Psychology, 24, 95-114.

*Kolb, D. A., Winter, S. K., & Berlew, D. E. (1968). Self-directedchange: Two studies. Journal of Applied Behavioral Science, 4,453-471.

*Latham, G. P., & Steele, T. P. (1983). The motivational effects ofparticipation versus goal setting on performance. Academy ofManagement Journal, 26, 406-417.

*Lee, T. W., Locke, E. A., & Phan, S. H. (1997). Explaining theassigned goal-incentive interaction: The role of self-efficacy andpersonal goals. Journal of Management, 23, 541-559.

*Lese, K. P., & Robbins, S. B. (1994). Relationship between goalattributes and the academic achievement of Southeast Asianadolescent refugees. Journal of Counseling Psychology, 41,45-52.

Locke, E. A. (1968). Toward a theory of task motivation andincentives. Organizational Behavior and Human Perfor-mance, 3, 157-189.

*Locke, E. A., Fredrick, E., Buckner, E., & Bobko, P. (1984).Effect of previously assigned goals on self-set goals and per-formance. Journal of Applied Psychology, 69, 694-699.

*Locke, E. A., Fredrick, E., Lee, C., & Bobko, P. (1984). Effect ofself-efficacy, goals, and task strategies on task performance.Journal of Applied Psychology, 69, 241-251.

Locke, E. A., & Latham, G. P. (1990). A theory of goal setting andtask performance. Englewood Cliffs, NJ: Prentice Hall.

Locke, E. A., Latham, G. P., & Erez, M. (1988). The determinantsof goal acceptance and commitment. Academy of ManagementReview, 13, 23-39.

Locke, E. A., Shaw, K. N., Saari, L. M., & Latham, G. P. (1981).Goal setting and task performance: 1969-1980. PsychologicalBulletin, 90, 125-152.

*Locke, E. A., Smith, K. G., Erez, M., Chah, D., & Schaffer, A.(1994). The effects of intra-individual goal conflict on perfor-mance. Journal of Management, 20, 67-91.

*Martocchio, J. J., & Dulebohn, J. (1994). Performance feedbackin training: The role of perceived controllability. PersonnelPsychology, 47, 357-373.

*Mathieu, J. E. (1992). The influence of commitment to assigned

goals and performance on subsequent self-set goals and perfor-mance. Journal of Applied Social Psychology, 22, 1012-1029.

*Matsui, T., Kakuyama, T., & Onglatco, M. U. (1987). Effects ofgoals and feedback on performance in groups. Journal of Ap-plied Psychology, 72, 407-415.

*McCaul, K. D., Hinsz, V. B., & McCaul, H. S. (1987). Theeffects of commitment to performance goals on effort. Journalof Applied Social Psychology, 17, 437-452.

*Mento, A. J., Cartledge, N. D., & Locke, E. A. (1980). Marylandvs. Michigan vs. Minnesota: Another look at the relationship ofexpectancy and goal difficulty to task performance. Organiza-tional Behavior and Human Performance, 25, 419-440.

Mento, A. J., Locke, E. A., & Klein, H. J. (1992). The relationshipof goal level to valence and instrumentality. Journal of AppliedPsychology, 77, 395-405.

Mento, A. J., Steel, R. P., & Karren, R. J. (1987). A meta-analyticstudy of the effects of goal setting on task performance: 1966-1984. Organizational Behavior and Human Decision Pro-cesses, 39, 52-83.

*Mone, M. A., & Baker, D. D. (1989). Stage of task learning as amoderator of the goal-performance relationship. Human Perfor-mance, 2, 85-99.

*Oldham, G. R. (1975). The impact of supervisory characteristicson goal acceptance. Academy of Management Journal, 18, 461-475.

*Oliver, R. L., & Brief, A. P. (1983). Sales managers goal com-mitment correlates. Journal of Personal Selling and Sales Man-agement, 3, 11-17.

*Orbach, I., Iluz, A., & Rosenheim, E. (1987). Value systems andcommitment to goals as a function of age, integration of per-sonality, and fear of death. International Journal of BehavioralDevelopment, 10, 225-239.

*Organ, D. W. (1977). Intentional vs. arousal effects of goalsetting. Organizational Behavior and Human Performance, 18,378-389.

Phillips, J. M., Hollenbeck, J. R., & Ilgen, D. R. (1996). Preva-lence and prediction of positive discrepancy creation: Examin-ing a discrepancy between two self-regulation theories. Journalof Applied Psychology, 81, 498-511.

*Podsakoff, P. M., MacKenzie, S. B., & Ahearne, M. (1997).Moderating effects of goal acceptance on the relationship be-tween group cohesiveness and productivity. Journal of AppliedPsychology, 82, 974-983.

*Racicot, B. M., Day, D. V., & Lord, R. G. (1991). Type Abehavior pattern and goal setting under different conditions ofchoice. Motivation & Emotion, 15, 67-79.

*Renn, R. W. (1998). Participation's effect on task performance:Mediating roles of goal acceptance and procedural justice. Jour-nal of Business Research, 41, 115-125.

*Riedel, J. A., Nebeker, D. M., & Cooper, B. L. (1988). Theinfluence of monetary incentives on goal choice, goal commit-ment, and task performance. Organizational Behavior and Hu-man Decision Processes, 42, 155-180.

*Roberson, L. (1989). Assessing personal work goals in the orga-nizational setting: Development and evaluation of the workconcerns inventory. Organizational Behavior and Human De-cision Processes, 44, 345-367.

*Roberson, L. (1990). Prediction of job satisfaction from charac-

896 KLEIN, WESSON, HOLLENBECK, AND ALGE

teristics of personal work goals. Journal of OrganizationalBehavior, 11, 29-41.

*Sabol, B. A. (1979). The relationship of goal acceptance, goalorientation and participation in goal setting to employee per-formance. Unpublished doctoral dissertation, Case Western Re-serve University, Cleveland, OH.

Sackett, P. R., Harris, M. M., & Orr, J. M. (1986). On seekingmoderator variables in the meta-analysis of correlational data: AMonte-Carlo investigation of statistical power and resistance toType-1 error. Journal of Applied Psychology, 71, 302-310.

*Sagie, A. (1996). Effects of leader's communication style andparticipative goal setting on performance and attitudes. HumanPerformance, 9, 51-64.

*Shalley, C. E., & Oldham, G. R. (1985). The effects of goaldifficulty and expected external evaluation on intrinsic motiva-tion: A laboratory study. Academy of Management Journal, 28,628-640.

*Shutz, P. A. (1993). Additional influences on response certitudeand feedback requests. Contemporary Educational Psychol-ogy, 18, 427-441.

*Smith, J. A., Hauenstein, N. M. A., & Buchanan, L. B. (1996).Goal setting and exercise performance. Human Performance, 9,141-154.

Staw, B. M. (1982). Counterforces to change. In P. S. Goodman &Associates (Eds.), Change in organizations (pp. 87-121). SanFrancisco: Jossey-Bass.

Stone, E. P., & Hollenbeck, J. R. (1984). Some issues associatedwith moderated regression. Organizational Behavior and Hu-man Performance, 34, 195-213.

*Theodorakis, Y. (1996). The influence of goals, commitment,self-efficacy and self-satisfaction on motor performance. Jour-nal of Applied Sport Psychology, 8, 171-182.

*Tubbs, M. E. (1993). Commitment as a moderator of the goal-performance relation: A case for clearer construct definition.Journal of Applied Psychology, 78, 86-97.

*Tubbs, M. E., & Dahl, J. G. (1991). An empirical comparison ofself-report and discrepancy measures of goal commitment. Jour-nal of Applied Psychology, 76, 708-716.

*Tziner, A., & Kopelman, R. (1988). Effects of rating format ongoal-setting dimensions: A field experiment. Journal of AppliedPsychology, 73, 323-326.

*Vance, R. J., & Collela, A. (1990). Effects of two types offeedback on goal acceptance and personal goals. Journal ofApplied Psychology, 75, 68-76.

* Vancouver, J. B. (1997). The application of HLM to the analysisof the dynamic interaction of environment, person and behavior.Journal of Management, 23, 795-818.

Wanous, J. P., Sullivan, S. E., & Malinak, J. (1989). The role of

judgment calls in meta-analysis. Journal of Applied Psychol-ogy, 74, 259-264.

*Weingart, L. R., & Weldon, E. (1991). Processes that mediate therelationship between a group goal and group member perfor-mance. Human Performance, 4, 33-54.

Whitener, E. M. (1990). Confusion of confidence intervals andcredibility intervals in meta-analysis. Journal of Applied Psy-chology, 75, 315-321.

*Whitney, K. (1994). Improving group task performance: The roleof group goals and group efficacy. Human Performance, 7,55-78.

Wofford, J. C., Goodwin, V. L., & Premack, S. (1992). Meta-analysis of the antecedents of personal goal level and of theantecedents and consequences of goal commitment. Journal ofManagement, 18, 595-615.

*Wofford, J. C., & Srinivason, T. N. (1983). Experimental tests ofthe leader-environment-follower interaction theory of leader-ship. Organizational Behavior and Human Decision Processes,32, 35-54.

Wood, R. E., Mento, A. J., & Locke, E. A. (1987). Task complex-ity as a moderator of goal effects: A meta-analysis. Journal ofApplied Psychology, 72, 416-425.

*Wright, P. M. (1989). Test of the mediating role of goals in theincentive-performance relationship. Journal of Applied Psychol-ogy, 74, 699-705.

*Wright, P. M., Hollenbeck, J. R., Wolf, S., & McMahan, G. C.(1995). The effects of varying goal difficulty operationalizationson goal setting outcomes and processes. Organizational Behav-ior and Human Decision Processes, 61, 28-43.

*Wright, P. M., & Kacmar, M. K. (1994). Goal specificity as adeterminant of goal commitment and goal change. Organiza-tional Behavior and Human Decision Processes, 59, 242-260.

*Wright, P. M., & Kacmar, K. M. (1995). Mediating roles ofself-set goals, goal commitment, self-efficacy, and attractive-ness in the incentive-performance relation. Human Perfor-mance, 8, 263-296.

*Wright, P. M., O'Leary-Kelly, A. M., Cortina, J. M., Klein, H. J.,& Hollenbeck, J. R. (1994). On the meaning and measurementof goal commitment. Journal of Applied Psychology, 79, 795-803.

*Yukl, G. A., & Latham, G. P. (1978). Interrelationships amongemployee participation, individual differences, goal difficulty,goal acceptance, goal instrumentality, and performance. Person-nel Psychology, 31, 305-323.

Received August 27, 1998Revision received January 5, 1999

Accepted January 6, 1999 •